"PSA AS A TOOL FOR EVALUATION OF AGEING EFFECTS ON THE SAFETY OF NPPs" Andrei Rodionov

1. "PSA AS A TOOL FOR EVALUATION OF AGEING EFFECTS ON THE SAFETY OF NPPs" Andrei Rodionov

2. 1. INTRODUCTION : PSA Application in the Long-Term Operation of NPPs Rationale of the Ageing PSA Network 2. SELECTION OF SYSTEMS, STRUCTURES AND COMPONENTS SENSITIVE TO AGEING Proposed Approach Related Uncertainties and Limitations 3. ACTIVE COMPONENTS RELIABILITY ASSESSMENT Task Specification: Incorporation of Ageing Effects into the PSA Model Time-Dependent Reliability Models Data issues 4. IMPACT OF AGEING EFFECTS ON THE SYSTEM AND PLANT LEVELS Approach Proposed for Demonstration System Level Effect Plant Level Effect 5. CONCLUSIONS. (+Summary of activities and future planning)

3. INTRODUCTION Motivations : • NPPs are getting older • Probabilistic safety goals have to be maintained for the LTO • Application of Risk Informed Regulation Questions : • Could PSA be applied to ageing assessments? • How realistically do PSA models reflect important ageing issues? • Are any modifications or revisions of PSA assumptions needed to apply a PSA approach to risk-informed decision-making with regard to ageing evaluation? • What data are available and how representative are they with regard to important ageing issues?

4. Background and motivations Actual situation worldwide: • 213 (49%) are between 20 and 30 years old, • 115 (26%) units are between 30 and 40 years old, In total : about 3/4 of the 438 reactors Long-term operation:on July 2006, 1/2 of the licensed NPPs in US had received or were under review for license renewal Situation in Europe (EU): • 88 (60%)reactors from 146 are in the age of 20-30 years old • 24 (16%) between 30 and 40 years old

5. Background and motivations VVERreactors inEU MS (19 reactors) : • about the same ages distribution. * RF -15 and UA – 15 reactors in operation • EU MS - Prevision for 2018 • 19 (14%)reactors from139need to be licensed for life extension (LTO) or shutdown, • 94 (67%)have to be under evaluation for life extension or definitive shutdown. **supposing that Sweden and Germany will continuer NPPs operation only up to the end of design lifetime (40 years) and 4 new reactors will be commissioned.

6. Traditional analysis PSA 1. Define change 2. Perform engineering analysis 3. Define implementation / monitoring program 4. Submit a proposal background and motivations • Probabilistic Safety Goal (INSAG-12) : “The target for existing nuclear power plants consistent with the technical safety objective is a frequency of occurrence of severe core damage that is below about 10–4 events per plant operating year. Severe accident management and mitigation measures could reduce by a factor of at least ten the probability of large off-site releases requiring short term off-site response.” • Risk Informed Regulation

7. Background and motivations • Ageing effect on unit/SSC reliability and safety • Main activities related to the ageing management : • Ageing management • Life time extension • Surveillance and Maintenance optimisation

8. Background and motivations Current situation • Standard PSA tools do not adequately address important ageing issues as : • assumption on component reliability model l=const, • reliability data may not adequately represent the current status of the plants, • existing PSAs overlook some components (e.g. cables, structures, etc.) as having a very low failure probability, but they may have increasing weight due to ageing effects • Limited experience with applying of PSA approach to evaluate ageing effects : • no commonly accepted methods, • all the studies are performed by relatively isolated organisations, • and publications on the subject are scarce

9. Needs for Advances Time-Dependent Reliability Analysis • Possible impact to PSA results • Passive components • IE frequencies and definitions (high Risk Importance) • Safety Systems unavailability due to the failures of non-redundant parts • Active components • Safety Systems unavailability due to the ageing failures, • Safety Systems unavailability due to the unplanned maintenance, • IE frequencies (when modeled with a FT), • Intersystem CCF • Practical implications Risk Profile and RI factors used as decision criteria in many applications could be changed in time. These changes have to be considered in decision making process.

10. EC JRC Network on Aging PSA : goals, tasks, working approach JRC Ageing PSA Network Objectives : Using common resources of Network participants to identify, develop and demonstrate methods and approaches which could help PSA developers and users • to promote the use of PSA for ageing management and risk-informed applications of Nuclear Power Plants in LTO, • to incorporate the effects of equipment ageing into current PSA models to perform engineering analysis, • in case where age-dependent PSA couldn’t be applied (absence or non-adequacy of ageing probabilistic model, lack of data, etc.), to specify and prioritize reliability monitoring actions/approach to assure that potential decreasing of reliability of SSC would be identified and corrected in time

11. EC JRC Network on Aging PSA : goals, tasks, working approach Main tasks : (Terms of references EUR 22645 EN ) • Task 1. Organization and coordination of network activities • Task 2. Analysis of main PSA tasks with regards to Ageing PSA (in progress) • Task 3. Selection of the SSC to be considered in Ageing PSA (in progress) • Task 4. Reliability and data analysis for active components (in progress) • Task 5. Reliability and data analysis for active components II. Common Cause Failures (started) • Task 6. Reliability and data analysis for passive components (started) • Task 7. Incorporation of age-depended reliability parameters and data into PSA model (in progress) • Task 8. Ageing PSA development and applications (started)

12. EC JRC Network on Aging PSA : goals, tasks, working approach Participants : • 15 organizations from EC MS, Swiss, RF, Armenia and Korea (network agreement). In addition, active participation from CNSC/Canada, D.Kelly and C.Atwod/USA Network Operation : • Working groups • Case studies and benchmark exercises, • Steering Committee, • Expert meetings, • Workshops • Web site : http://safelife.jrc.nl/APSA

13. 1. INTRODUCTION : PSA Application in the Long-Term Operation of NPPs Rationale of the Ageing PSA Network 2. SELECTION OF SYSTEMS, STRUCTURES AND COMPONENTS SENSITIVE TO AGEING Proposed Approach Related Uncertainties and Limitations 3. ACTIVE COMPONENTS RELIABILITY ASSESSMENT Task Specification: Incorporation of Ageing Effects into the PSA Model Time-Dependent Reliability Models Data issues 4. IMPACT OF AGEING EFFECTS ON THE SYSTEM AND PLANT LEVELS Approach Proposed for Demonstration System Level Effect Plant Level Effect 5. CONCLUSIONS

14. SELECTION OF SSCs SENSITIVE TO AGEING • Current Ageing Management practice : • often limited to passive components • if not based on expert judgments • Proposed Approach includes three elements : • prioritize the SSCs which are modeled in PSA using RIF, • perform trend analysis of available reliability data, • use qualitative assessment or Ageing Failure Modes and Effect Analysis (AFMEA) for a limited number of components All three steps are consecutive and complementary.

15. Component modeled in PSA Risk Importance Reliability data Reliability data Reliability data Trend analysis AFMEA Trend analysis AFMEA Select Rank 6 Select Rank 7 No No Select Rank 3 Select Rank 5 No No Trend analysis AFMEA Select Rank 1 Select Rank 2 AFMEA No Select Rank 4 No Selection process

16. SELECTION OF SSCs SENSITIVE TO AGEING

17. SELECTION OF SSCs SENSITIVE TO AGEING Related Uncertainties and Limitations Prioritization by risk importance limitations: • (+) permits significantly reduce the list of SSCs, • (-) it deals only with SSCs modeled in PSA, • (-) importance of particular SSCs could be changed with time due to the ageing, Trend analysis • (+) direct indication of ageing, • (+) provide the basis for modelling, • (-) lack of data, Qualitative assessment • (+) could cover SSCs initially neglected in PSA and those for which there is not enough failure data for trend analysis, • (-) very time-consuming and resource-intensive, • (-) qualitative results

18. 1. INTRODUCTION : PSA Application in the Long-Term Operation of NPPs Rationale of the Ageing PSA Network 2. SELECTION OF SYSTEMS, STRUCTURES AND COMPONENTS SENSITIVE TO AGEING Proposed Approach Related Uncertainties and Limitations 3.ACTIVE COMPONENTS RELIABILITY ASSESSMENT Task Specification: Incorporation of Ageing Effects into the PSA Model Time-Dependent Reliability Models Data issues 4. IMPACT OF AGEING EFFECTS ON THE SYSTEM AND PLANT LEVELS Approach Proposed for Demonstration System Level Effect Plant Level Effect 5. CONCLUSIONS

19. l(t) lav, i [ti, ti+1] t ACTIVE COMPONENTS RELIABILITY ASSESSMENT Integration of SSC ageing model to PSA • time-dependent reliability parameters • PSA model • CDF at different age-points (10, 20, 30 and 40 years of operation) Time-dependent reliability model : • collect and process reliability data, • fit one of the parametrical models and calculate the model parameters, • if necessary, improve the results byBayesian approach and generic data, • evaluate the unavailability factors for given age values, taking into account model parameters, periodic tests and maintenance data

20. Time-Dependent Reliability Models Choose the model • Constant failure intensity (rate) : = const • Linear : • Log-linear or exponential : • Power-low (Weibull) : q2> 0 means a positive trend in time The procedure for choosing the model and parameters includes the following steps: • verification of model validity, • parameter estimation, • characterization of uncertainties of estimated parameters and the whole model, • assessment of possible extrapolation and uncertainties of extrapolation

21. Time-Dependent Reliability Models The following issues require specific attention when analyzing and interpreting the results: • relative increase in failure intensity (rate) in time with regard to constant failure rate, • impact of burn-in failures, • uncertainties of extrapolation

22. Time-Dependent Reliability Models

23. Data issues Possible sources of data : • reliability data for NPPs (generic and/or specific), • reliability data for similar components from other industries, • accelerated ageing reliability tests Reliability data from fossil plants 20-40 years old (NERC and VGB data) Main conclusions In general (good news), with age plants “learn” : availability and performance increase or maintained constant with the age of the plant Particular cases of ageing : • increasing failure rates are identified for several types of components as pumps, funs, valves, switchers, boilers, heaters, etc. ( ageing parameters for linear aging model are comparable with TRIGALEX data ); • “minimal repair” maintenance strategy could lead to increase in failure rate; • even with decreasing or constant failure frequency, reliability of component could degrade with age because of increasing of repair frequency and its duration

24. Activities and results : Reliability and data analysis for active components (Task 4)

25. PSA Reliability DB Raw data sources Component reliability parameters Operating (defect) and maintenance logs IE frequencies Abnormal Operational Events (LER) Reporting System Processed component failure, performance and maintenance data Operating and maintenance procedures Processed Operational Event data Design, commissioning and manufacturing information Generic data (Reliability parameters and IE frequencies) Other reliability DBs (vendor DB, NDE of piping systems DB, I&C elements DB, etc.) Data issues Reliability data for NPPs: (questionnaire on data availability and accessibility - conclusions EUR23084EN) PSA reliability data • well structured and of high quality, • easily accessible (DB or electronic docs), • not sufficient for ageing assessments (l=const => n + St) Additional data has to be extracted and processed from raw data

26. Data issues Processed data (2b) Nomenclature of PSA Component Reliability Data Data for parameters estimation 1. Component group 2.Component description / limits 3. Operating mode 4. Component function 5. Failure modes 6. Failure criteria 7. Sample (# and list of components) 8. Observation period 9. Stressors (av. # of hrs or d per year) 10. Cumulated stressors 11. # of failures per failure mode 12. Estimation method / assumptions Data for cumulated operation time 13. Component ID (sub-group) 14. # of compon. in sub-group 15. # of hrs or demands / year 16. Observation period 17. Total cumulated stressors • Reliability parameters (2a) • 1. Component group • 2. Failure modes • 3. Parameters : • Failure rate, • Failure prob. per demand, • Mean time to repair, • Unavailability due to • the maintenance. • 4. Uncertainty Failure data 18. Unit / component ID 19. Failure date 20. Reactor state 21. Failure mode 22. Criticality factor 23. Repair time 24. Failure cause / description

27. Data issues Age-dependent reliability models : • simple models or trend assessment; • models which include test and maintenance evaluations; • comprehensive models Categories of additional data needed: Type 1 models • component commissioning date, • failure/censoring times, • component replacement date. Type 2 models - data listed for Type 1, plus: • tests and maintenance strategy – type and periodicity, • degree of component renewal (CM and PM) In fact these data needed for T1 models to make the assumptions about renewal process Type 3 - data listed for Type 1 and Type 2, plus: • component lifetime, • cumulative number of hours in operation, number of demands, • average and extreme levels of operating and environmental stressors

28. Data issues Main conclusions : • Processed data about failures and component performance could be certainly used for age-dependent reliability analysis, but is not enough for this purpose • PSA reliability data collection process does not include any requirement to perform a statistical validation of assumptions about constant failure rate or trend analysis • Improving reliability data collection could greatly help with PSA and age-dependent reliability analysis applications in risk-informed decision making process

29. Needs for Advances Time-Dependent Reliability Analysis • Practical implications : • reliability analysis of ageing trend has to be a part of the process. • Experience of application • analysis of ageing trends (large populations) : • (?) data collection covers only risk important components => no data available for all components potentially susceptible to ageing; also for many components statistical data are not enough, • Typical results of analysis : • ~ 30% of analysis cases : not enough failure statistic (with sufficient demands/operating times) = very reliable, no ageing • ~ 20-25% of analysis cases : constant failure rate = no ageing, • ~ 5-15% of analysis cases : decreasing failure rate – after burn-in period and after modifications • What have to be considered as parameters? • ~ 20-30% of analysis cases : increasing failure rates – detailed engineering investigation is required before conclusion about ageing • ??? issues

30. 1. INTRODUCTION : PSA Application in the Long-Term Operation of NPPs Rationale of the Ageing PSA Network 2. SELECTION OF SYSTEMS, STRUCTURES AND COMPONENTS SENSITIVE TO AGEING Proposed Approach Related Uncertainties and Limitations 3. ACTIVE COMPONENTS RELIABILITY ASSESSMENT Task Specification: Incorporation of Ageing Effects into the PSA Model Time-Dependent Reliability Models Data issues 4. IMPACT OF AGEING EFFECTS ON THE SYSTEM AND PLANT LEVELS Approach Proposed for Demonstration System Level Effect Plant Level Effect 5. CONCLUSIONS

31. IMPACT ON THE SYSTEM AND PLANT LEVELS • To calculate the impact of ageing on the risk profile as a function of unit age, it was proposed to use CDF as the average value at one-year intervals calculated for different age points, for example for 10, 20, 30 and 40 years in operation • A three-loop PWR PSA model for a Large LOCA initiating event was considered for this purpose. The model consists of 4 Event Trees developed for full power operation and hot shutdown reactor states • The set of “virtual” reliability data was prepared on the basis of the results of case studies, available generic data sources and expert opinions. The data includes time-dependent reliability models for certain mechanical, electrical and I&C components of Low Pressure Safety Injection (LPSI) and Containment Spray (CSS) Systems • “As bad as old” preventive maintenance was considered in all cases • Quantification has been calculated for a reference value (no ageing effects considered) and age points of 10, 20, 30 and 40 years

32. Examples of component reliability data

33. System Level Effect Results of calculations for CSS: • Unavailability increases with time by > than one order of magnitude with regard to the reference value (see Fig. 3) • Up to 30, the main contributor to unavailability is the failure of level sensors in the RWST But at the age of 40 the dominant impact on system unavailability is failure of the CSS pump motors • “No ageing” contributors : HF - human error (20% fractional contribution to the reference value) and CSS pumps fail to start (2% contribution to the reference value)

34. Plant Level Effect Impact of SSC ageing on CDF • CDF increases from 6.58E-8 at 10 years to 6.19E-7 at 40 years. In comparison with the reference value the increase is by a factor of 8.6 by the end of the designed lifetime • Once the most sensitive components (LPSI and CSS pump motors and level sensors) are in the MCSs of dominant sequences, the relative contribution of the sequences to the total risk of Large LOCA remains approximately the same with age • The same picture can be seen for contributions to the risk associated with the different reactor states and location of the pipe break

35. Plant Level Effect • Sensitivity analysis of the reliability model chosen for the pump motors • As a base case, the log-linear model was considered. The Weibull model was applied as alternative for sensitivity analysis • The generic conclusion from this analysis is the need to examine the accuracy of several model alternatives before applying one to PSA

36. 1. INTRODUCTION : PSA Application in the Long-Term Operation of NPPs Rationale of the Ageing PSA Network 2. SELECTION OF SYSTEMS, STRUCTURES AND COMPONENTS SENSITIVE TO AGEING Proposed Approach Related Uncertainties and Limitations 3. ACTIVE COMPONENTS RELIABILITY ASSESSMENT Task Specification: Incorporation of Ageing Effects into the PSA Model Time-Dependent Reliability Models Data issues 4. IMPACT OF AGEING EFFECTS ON THE SYSTEM AND PLANT LEVELS Approach Proposed for Demonstration System Level Effect Plant Level Effect 5. CONCLUSIONS

37. Conclusions • Ageing effects can alter the conclusions of reference PSA studies. In particular, they can impact on: • system unavailability and CDF, • dominant accident sequences and contributors to CDF, • component risk importance measures. • Considering ageing effects in PSA and reliability analysis could help in the selection and prioritization of SSCs for ageing management and maintenance optimization as part of a risk-informed decision-making process • The main problems relate to the methodology, data and resource availability • The purpose of the EC-JRC Ageing PSA Network’s activities is to provide PSA engineers with practical approaches, methods and advice on how evaluate the importance of ageing issues by means of PSA modeling. The results presented demonstrate methods and approaches proposed for the selection of SSCs susceptible to ageing, the development of time-dependent reliability models and the evaluation of ageing effects on overall plant safety

38. Summary of 2008 activities • Scientific publications : • 1 published + 2 submitted (Task 4) • Case Studies : 2 completed • Task 3 EUR report is ready for registration • Task 4 EUR 23079 EN is published • End users support : • Guideline for Analysis of Data Related to Aging of NPP Components and Systems – final version is ready for the review (JRC reference report, Task 4) • Collaboration in ISTC project on APSA for Armenian NPP • Policy Support : • E&I Training on Advanced Time-Dependent Reliability Data Analysis, 6-9 October 2008 • Contribution as Invited expert to IAEA workshop on VVER PSA • Results dissemination : • PSA’2008 Int. Conference – 6 papers presented by Network participants • interview for EUROSAFE Tribune • Update of APSA Web-site

39. Conclusions The following topics are planned for further development : • areas of possible PSA application in ageing management via risk-informed approaches (Task 8), • reliability data collection and parameter estimation : CCF (Task 5), • passive component age-dependent reliability models (Task 6), • the incorporation of ageing effects into the PSA model (Task 7). Network Meeting December 4-5, 2008, Prague, Czech Republic.

40. Thank you very much for your attention!